bg-curve-remover

A pure JavaScript background remover for browser workflows.

bg-curve-remover removes image backgrounds using an in-house, browser-side extraction pipeline:

• border color modeling
• background flood-growth segmentation
• curve/edge alpha softening
• small artifact cleanup

Install

npm install bg-curve-remover

Quick Start (Browser)

import { removeBackgroundToObjectURL } from 'bg-curve-remover';

const input = document.querySelector('#file').files[0];
const resultUrl = await removeBackgroundToObjectURL(input);
document.querySelector('#preview').src = resultUrl;

Core API

removeBackground(input, options?)

Removes the background and returns a processed image object.

• Browser usage: returns a Blob (PNG by default)
• Input types: File, Blob, URL string

import { removeBackground } from 'bg-curve-remover';

const outputBlob = await removeBackground(file, {
  output: { type: 'image/png', quality: 0.98 },
  extraction: { edgeSoftness: 26, minSubjectRegion: 120 }
});

removeBackgroundToObjectURL(input, options?)

Convenience wrapper for browser previews.

• Calls removeBackground(...)
• Converts result Blob to URL.createObjectURL(...)

import { removeBackgroundToObjectURL } from 'bg-curve-remover';

const url = await removeBackgroundToObjectURL(file);
img.src = url;

Options

output

Output image configuration.

• type: MIME type (recommended: image/png for transparency)
• quality: compression quality (used where relevant)

{ output: { type: 'image/png', quality: 0.98 } }

extraction

Controls your in-house subject extraction pipeline.

• borderSampleStep (default 2): border sampling stride for background color model
• bgDistancePercentile (default 0.35): base threshold from color-distance distribution
• bgGrowMultiplier (default 1.25): expansion factor when flood-filling background
• minSubjectRegion (default 120): remove tiny foreground islands
• edgeSoftness (default 26): feathering amount on extracted edges

{
  extraction: {
    borderSampleStep: 2,
    bgDistancePercentile: 0.35,
    bgGrowMultiplier: 1.25,
    minSubjectRegion: 120,
    edgeSoftness: 26
  }
}

This helps remove leftover stray pixels both outside and inside the subject region while preserving smoother edges.

Demo

Live demo (GitHub Pages):

• https://akmalnawfer.github.io/bg-curve-remover/

Run demo locally:

npm install
npm run demo

Use in Your Project

Typical flow:

1. User selects or drops image
2. Call removeBackground(...)
3. Show preview, upload output, or download PNG

Example (download):

const resultBlob = await removeBackground(file, {
  output: { type: 'image/png', quality: 0.98 }
});

const url = URL.createObjectURL(resultBlob);
const a = document.createElement('a');
a.href = url;
a.download = 'cutout.png';
a.click();
URL.revokeObjectURL(url);

Notes

• First run may be slower while model assets are loaded.
• PNG output is recommended for transparent backgrounds.
• Complex hair/fur/translucent edges can still benefit from manual touch-up depending on source image quality.
• This package does not call your backend services; extraction runs locally in-browser.

License

MIT